Consider taking a snapshot of an index I and mounting it as M-I. Now if a snapshot of M-I is taken, it will be an independent snapshot, meaning that the segment files will be re-copied to the snapshot repository, instead of using the segment files from I. Note that if M-I needs to be recovered, or is restored, the segment files will be copied from the snapshots of I, not even using the snapshot of M-I. This effectively doubles the storage of mounted indices in the snapshot repository, with half that space being pure waste.